Bacteria carry many types of obligate intracellular parasites, including plasmids and bacteriophage (viruses that infect bacteria). During infection, these parasites redirect intracellular resources away from bacterial processes toward parasite production. Because parasite-induced metabolic changes influence host phenotypes, parasitic infection is predicted to alter how microbes contribute to important community and ecosystem functions. Yet little is known about how infection shapes interactions between host and non-host species. Here we integrate a genome-scale metabolic modeling approach with an in vitro obligate cross-feeding system to investigate the metabolic consequences of two intracellular parasites of Escherichia coli – the F-plasmid and the filamentous phage M13 – on interactions between bacteria in a multispecies community composed of E. coli, Salmonella enterica, and Methylorubrum extorquens. Modeling predicts that metabolic conflict due to parasitic infection of E. coli changes the types and quantities of compounds secreted by infected hosts, increasing species evenness when bacterial species are engaged in obligate mutualism. These theoretical results are corroborated by in vitro experiments. Our work emphasizes that microbes infected by diverse obligate intracellular parasites have fundamentally different metabolisms from uninfected cells and demonstrates that these metabolic shifts can have significant consequences for microbial community structure and function.

The immune system works as a well-oiled machine during a pathogen attack. The dysfunction of this well-oiled machine can be catastrophic, such as sepsis, with death being one of the ultimate consequences. Although several studies have shown the progression of sepsis and the complex dynamic interaction between the immune and metabolism playing a role, the prediction of the onset of sepsis and sepsis shock is still challenging. The intellectual construct of sepsis is based on this immune dysfunction either in its failure to clear out the foreign invaders attacking the body or the breakdown of the counter-attacking immune system due to internal failures of their mechanisms. The current work here proposes an ODE-based model that accounts for immune, metabolic, and physiological variables to address this issue of the onset of sepsis. The model developed was used to analyze system-level dynamics and predict the healthy and sepsis response. The interplay between pro-inflammatory and anti-inflammatory cytokines and Nitric oxide and lactate levels form the basis of sensitivity toward sepsis. Sensitivity Analysis gives an understanding of the different stages of the infection and the various parameter regimes that progress to sepsis and shock. The model makes use of states projected along with parameter regimes for the ICU patients with the model tweaked for the ICU patients to match the onset of sepsis and make predictions for its progression thus creating a model-based control system for the prediction of the onset of sepsis and septic shock in ICU patients.

Cancer metastasis accounts for nearly 90% of cancer-related deaths. Epithelial to Mesenchymal Transition (EMT) of cancer cells is key to cancer metastasis. Recent preclinical and clinical data suggest that hybrid E/M cell states harbor the highest metastatic fitness instead of the fully mesenchymal cells. However, the mechanistic details of their survival strategies during metastasis remain unclear. We model the dynamics of a minimalistic gene regulatory network (GRN) between regulators of EMT and PD-L1, a known immune suppressor. We show that hybrid E/M states are highly likely to exhibit high PD-L1 levels, like those seen in mesenchymal cells, thus obviating the need to undergo a complete EMT to cause immunosuppression. We show that the switch from an epithelial/low-PDL1 state to a hybrid/mesenchymal state with high-PDL1 is reversible. We demonstrate that acquiring phenotypic resistance to targeted therapy can co-occur with high PD-L1 levels, enabling cross-resistance and enhancing breast cancer cell fitness during metastasis. We validate our model predictions by extensive analysis of transcriptomic datasets across multiple cancers at bulk and single-cell levels. Our results highlight how the emergent dynamics of interconnected GRNs can coordinate various axes of cellular fitness during metastasis, thus laying the foundation for the rational design of clinical therapies.

Cancer remains a major challenge in the field of medicine, necessitating innovative therapeutic strategies. Mitogen-activated protein kinase (MAPK) signaling pathways, particularly Extracellular Signal-Regulated Kinase 1 and 2 (ERK1/2), play pivotal roles in cancer pathogenesis. Recently, ERK5 (also known as MAPK7) has emerged as an attractive target due to its compensatory role in cancer progression upon termination of ERK1 signaling. This study explores the potential of Compound 22ac, a novel small molecule inhibitor, to simultaneously target both ERK1 and ERK5 in cancer cells. Using molecular dynamics simulations, we investigate the binding affinity, conformational dynamics, and stability of Compound 22ac when interacting with ERK1 and ERK5. Our results indicate that compound 22ac forms strong interactions with key residues in the ATP-binding pocket of both ERK1 and ERK5, effectively inhibiting their catalytic activity. Further, the simulations reveal subtle differences in the binding modes of Compound 22ac within the two kinases, shedding light on the dual inhibitory mechanism. This research not only elucidates a structural mechanism of action of Compound 22ac, but also highlights its potential as a promising therapeutic agent for cancer treatment. The dual inhibition of ERK1 and ERK5 by Compound 22ac offers a novel approach to disrupting the MAPK signaling cascade, thereby hindering cancer progression. These findings may contribute to the development of targeted therapies that could improve the prognosis for cancer patients.

Elementary flux modes (EFMs) are minimal sets of reactions that can sustain steady state flux in a metabolic flux network. First proposed by Schuster and Hilgetag, any set of steady state fluxes in a metabolic network can be explained as a positive, linear combination of their EFM weights. This decomposition, however, is generally not unique, as there are typically exponentially many EFMs. We recently addressed this flux decomposition problem for the special case of unimolecular reaction networks involving single substrates and products. By imposing a Markovian constraint on EFMs, we modelled steady state particle fluxes as a cycle-history Markov chain (CHMC) to uniquely identify EFM weights that reconstruct any network fluxes. Here, we generalize our CHMC analysis to networks containing multispecies reactions. We propose to concept of atomic EFMs which are steady state pathways that trace the flow of individual atoms through a network. Using a state-of-the-art atom mapping algorithm, we enumerate carbon and nitrogen EFMs from four genome-scale networks and compute atomic EFM weights from a fifth HepG2 dataset. We find that enumerating atomic EFMs is computationally tractable compared to standard EFMs, and show how atomic EFMs characterize nutrient source metabolism through the network. Lastly, our analysis of the HepG2 atomic EFM weights reveals that the majority of steady state fluxes are explained by a small fraction of atomic EFMs. Our atomic CHMC method is a fast and novel tool to quantify metabolic remodelling as technologies to generate large-scale metabolic flux networks advance.

We propose Mitovolve as a statistical model that utilizes single-cell sequencing data to infer the evolutionary history of somatic mitochondrial DNA (mtDNA) mutations. Mitovolve models the evolution of a mtDNA mutation from one cell of origin with a specified number of mutant and wild-type mitochondrial genomes to its observed abundance across the numerous cells represented in single-cell sequencing datasets. The distribution of the number of mutant mitochondrial genomes in each daughter cell is derived by hypergeometric sampling from a pool of duplicated mitochondrial genomes in the parent cell, a process that is reiterated for each successive generation of new cells. Given the initial abundance of mutant mitochondrial genomes, the model calculates an exact probability distribution for the number of mutated mitochondrial genomes per cell after a specified number of replicative generations in the presence or absence of selective pressure parameters represented as a cubic polynomial function of the mutant allele frequency (MAF). Maximum likelihood estimation is used to fit many models (varying the initial mutant allele fraction and number of replicative generations) with and without selection to the observed data. A likelihood ratio test is then used to compute confidence sets and p-values for the presence and nature of selective pressure operating on the mtDNA mutation. As proof of concept, we applied Mitovolve to single cell sequencing data from a primary pediatric leukemia sample and found statistically significant evidence in favor of selective pressure operating on a tumor-enriched mtDNA mutation. Mitovolve will soon be available as an R package.

Background and Motivation: Single-cell RNA sequencing (scRNA-seq) has significantly enhanced the resolution and scale at which transcriptomics can be studied. Despite these advancements, it has also led to an expansion in the volume of data generated by single-cell studies, further complicated by the bimodal distribution of gene expression and ‘dropout’ events. To address these intricacies, MAST, a versatile statistical framework in R, employs a hurdle model—a dual-component generalized linear model adept at characterizing the discrete occurrence of gene expression from the continuous spectrum of expression levels. While R continues to be an invaluable tool for detailed statistical analysis, Python’s strengths in handling large datasets, scalability, and data integration techniques make it a suitable platform for projects involving large-scale genomic data.
Method and Results: We propose developing a Python-based hurdle model that integrates logistic regression for zero inflation and a negative binomial model for expression quantification. Adjustments include a refined approach to managing zeros and leveraging Python’s extensive libraries for optimization and model fitting. To demonstrate the efficacy of our framework, we validated it by analyzing the Peripheral Blood Mononuclear Cells dataset. Preliminary comparisons of our results with those obtained from the MAST indicate a strong positive Spearman’s correlation (ρ = 0.731) with a highly significant p-value (1.43e-257), showcasing the potential of our approach to offer comparable analytical depth. Further optimizations are underway to enhance model accuracy and data handling.
Conclusion: Our Python-based model offers a powerful approach for identifying differentially expressed genes across conditions in large scRNA-seq datasets.

At least 50,000 women in the United States are expected to be diagnosed this year with ductal carcinoma in situ (DCIS), a type of breast precancer. To decrease the risk of DCIS progression to invasive tumours, breast-conserving lumpectomies or mastectomies are typically performed. However, patient treatment and follow-up strategies remain to be improved given significant DCIS recurrence rate despite surgery as well as growing concern that such aggressive interventions may be unnecessary in some cases. To complement the search for genetic and epigenetic markers of recurrence, we employ constraint-based metabolic modeling to identify metabolic signatures of DCIS recurrence as a means to improve patient outcome. We leverage bulk RNA-sequencing data from the Translational Breast Cancer Research Consortium to identify expression patterns at initial diagnosis distinguishing patients who subsequently experienced recurrence within the next 5 years (n=63) from those with no recurrence (n=95). We mapped these expression differences to the most recent human genome-scale metabolic reconstruction (Human1) and identified cellular metabolic tasks that were significantly up and downregulated in patient cohorts (for instance, fatty acid synthesis upregulation and phospholipid synthesis downregulation). We further applied deconvolution techniques to identify metabolic rewiring at cell type resolution. Through our cell type-specific map of metabolism in DCIS, we provide a more comprehensive view of the interaction between precancerous cells and the tumour microenvironment which may, in turn, provide a unique perspective into recurrence and the transition from DCIS to tumour invasion.

Lytic polysaccharide monooxygenases (LPMOs) are recently discovered oxidoreductases that can oxidatively cleave the glycosidic bonds found in carbohydrate polymers enabling the depolymerisation of recalcitrant biomasses. While most have explored the role of LPMOs in a (plant) biomass conversion/biofuel context, alternative roles have emerged such as in pathogenesis of communicable diseases. Various experimental studies on LPMOs from Vibrio cholerae, Paenibacillus larvae, Listeria monocytogenes, Pseudomonas aeruginosa and Lysobacter enzymogenes have shown involvement in pathogenesis and virulence roles. However, the molecular mechanism of pathogenesis remains unclear. Here, through systems biology and network analysis, we aimed to decipher the interplay between LPMOs and other bacterial proteins, utilizing data from the STRING database to construct interaction maps. Using Gene Ontology (GO) enrichment analysis, protein-protein docking, network perturbation analysis, we have attempted to decipher the pathogenic role of LPMOs. We focused on V. cholerae LPMOs, namely gbpA and VC_A0140. GO enrichment analysis of gbpA revealed extracellular localization, metalloendopeptidase activity, metal ion binding as molecular functions , and proteolysis as biological processes. The protein-protein docking was done using HADDOCK for two pairs, gbpA with hap and gbpA with VC_1952 and analyzed using PDBsum, where the presence of salt bridges and hydrogen-bonded interactions indicate strong interactions. Similarly, VC_A0140 showed that it is extracellularly located, ATP binding, Metal ion binding, and metalloendopeptidase for molecular function, and proteolysis for biological processes. This study can help understand LPMOs’ role in pathogenesis, for future computational studies, and identify targets for designing inhibitors to prevent bacterial pathogens from invading host cells.

A major challenge with analyzing spatially resolved single-cell data collected at multiple time points is temporal misalignment. That is, spatial coordinates captured at different time snapshots may represent different tissue regions. Despite technological efforts to address this issue, the inherent variability among different subjects hampers complete success. Analyzing such data using existing pipelines may lead to problems, inefficiencies, and some degree of inaccuracy. To mitigate these challenges, we propose SpatialDNet, which fuses network and spatio-temporal models to quantify cellular behavior within a cellular microenvironment. SpatialDNet takes both spatial location information of cells or spots and genomic data as input. It constructs a rectangular mesh over the spatial region, creates network nodes from these rectangles, and utilizes multidimensional gene expression data to connect the nodes, forming a comprehensive network. We validated the analysis framework using a CODEX dataset obtained from the study of advanced melanoma tumors to estimate spatio-temporal variable genes, determine gene regulatory networks, and annotate the trajectory of the tumor-immune cell network. In the experimental protocol, CD8+ T cells activated ex-vivo with gp100 antigen and IL-2 were transferred into mice with B16-F10 tumors. Tumors were harvested and imaged at day 0, 1, 3, 5, and 12 post-treatment. The results of the analysis revealed distinct dynamic tumor-immune relationships following T-cell therapy. Specifically, there was a significant increase in tumor and immune cell proliferation, which continued until day 5 but declined by day 12. These findings could provide valuable insights into the efficacy of T-cell therapy among patients.

Computational modelling of cell state transitions has been a great interest of many researchers in the field of developmental biology, cancer biology and cell fate engineering. It allows us to study the biological systems and perform in-silico perturbations that would otherwise be expensive to do experimentally. Recent advancement in single-cell RNA sequencing (scRNA-seq) technology allows us to capture high-resolution snapshots of cells transitioning states across trajectories. Using these high-throughput datasets, we can train computational models that generate in-silico synthetic cells to faithfully mimic the developmental processes. Here we present OneSC, a platform that can simulate synthetic cells across developmental trajectories using systems of stochastic differential equations govern by a core transcription factors (TFs) regulatory network. Different from the current network inference methods, OneSC prioritizes on generating Boolean network that produces faithful cell state transitions and steady cell states that mimic real biological systems. Applying OneSC to real data, we inferred a core TF network using a mouse myeloid progenitor scRNA-seq dataset and showed that the dynamical simulations of that network generate synthetic single-cell expression profiles that faithfully recapitulate the four myeloid differentiation trajectories going into differentiated cell states (erythrocytes, megakaryocytes, granulocytes and monocytes). Finally, through the in-silico perturbations of the mouse myeloid progenitor core network, we showed that OneSC can accurately predict cell fate decision biases of TF perturbations that closely match with previous experimental observations. In summary, OneSC is a computational platform that simulates synthetic cells to mimic cell state transitions observed in real single-cell expression data.

Background: Functional cure in chronic hepatitis B (CHB) is defined as persistent loss of detectable serum hepatitis B surface antigen (HBsAg) and HBV DNA. Integrated HBV DNA (intDNA) in hepatocytes is a major source of HBsAg, but current methods to quantify intDNA require invasive liver biopsies.
Methods: We developed a mathematical model to estimate the relative abundance of hepatocytes with intDNA or covalently closed circular DNA (cccDNA) using serum viral measurements. This non-linear mixed-effects model was fit using stochastic approximate expectation maximum (SAEM) to data from CHB patients (n=119) in a randomized, double-blind, placebo-controlled study of Tenofovir Disoproxil Fumarate
(TDF) 300 mg/day or placebo over 3 years (NCT01522625). Paired core liver biopsies (n=42) collected per protocol at baseline and year 3 were used to determine observational relative cccDNA and intDNA content and for a novel comparison to modeled cccDNA and intDNA.
Results: Fitted model results provide individual-specific estimates of viral protein production rates and infection cell quantities at each collection time. There are significant correlations between modeled and observed cccDNA cell population (Pearson’s r=.40, p<.001) as well as intDNA cell population (Pearson’s r=.44, p<.001). Resulting viral protein rates indicate most sub-viral particles are produced by intDNA, in-line with biological expectation for this HBeAg- population.
Conclusions: Model results reflect biological expectation and can reproduce liver biopsy-derived results from only serum measurement. These results provide evidence for the utility of modeling of HBV infection dynamics in understanding cccDNA and intDNA disease burden through non-invasive means for HBeAg- CHB patients.

Transcriptomic integration with genome-scale metabolic models is crucial for deepening our understanding of complex systems in biology. While existing validation studies for transcriptomic integration offer some insights, their scope is limited, particularly for organisms like cyanobacteria, for which little metabolic flux data are available. The impact of pre-processing decisions on integration has scarcely been assessed beyond human models, with no thorough exploration of parameter choices in valve-based integration methods. This study evaluates these methodological decisions using the genome-scale model for Synechocystis sp. PCC 6803 (iSynCJ816 [1]) with existing transcriptomic data to explore their structural impact on the metabolic network and their effect in biomass-optimised scenarios. Preliminary analyses indicate that selecting an appropriate integration method requires consideration for the intended application, highlighting the need for tailored strategies that capture the desired biological insight. Upon completion of this study, we expect to establish foundational guidelines for accurately modelling the cellular phenotype, suitable for downstream analysis and broadly applicable across systems biology.
[1] Joshi, C.J., Peebles, C.A., & Prasad, A. (2017). Modeling and analysis of flux distribution and bioproduct formation in Synechocystis sp. PCC 6803 using a new genome-scale metabolic reconstruction. Algal Research-Biomass Biofuels and Bioproducts, 27, 295-310.

Modeling the interaction between the target protein and the inhibitor and analyzing the binding energy to understand the structure-activity relationship of the inhibitor is very helpful in new drug design research. In particular, recent improvements in computing power and advances in analysis equipment technology such as X-ray crystallography have made accurate structure prediction and rapid calculation possible. Accordingly, new drug development using computers is receiving attention.
Design of kinase inhibitors targeting the BCR-ABL oncoprotein is an effective strategy for the treatment of chronic myeloid leukemia (CML). In this study, we conducted a study to develop a pan-BCR-ABL inhibitor containing the imatinib resistance mutation T315I. To explain the experimentally observed Abl kinase inhibition results, molecular docking of indazole-based derivatives against the catalytic kinase domain of AblWT (PDB code: 3OXZ) and AblT315I (PDB code: 3OY3) was performed from a three-dimensional structural perspective. In particular, structural differences in the candidate substances that affect the Bcr-Abl kinase inhibitory effect were revealed through analysis of binding method, binding energy, and ligand-protein binding.

Motivation: Swiss-Prot keyword searches provide a wealth of biocurated proteins with which to gauge a topic. We provide a paradigm for directed investigation of KEGG pathway analysis, beginning with discrete correlate summation (DCS) clustering of Swiss-Prot sets. Dopamine/dopaminergic and angiotensin/renin related pole proteins assemble in pathway analyses upon a field of proteins also associated with 16 other designated keywords to provide a juxtaposed trial of their biological indices.

Our study focuses on generating a lightweight k-mer model tailored for RNA004 chemistry in nanopore sequencing, addressing challenges in accurate alignment and signal simulation, particularly in the absence of optimized official models. Leveraging the move table and k-mer sequences, our approach ensures precise event alignment from raw signals to basecalled sequences. We showcase the effectiveness of our custom k-mer model through high alignment rates (97.48%) and improved signal simulation accuracy compared to default models. Additionally, our 5-mer model exhibits similar performance as the default 9-mer models in methylation detection in m6A site detection. This work underscores the significance of lightweight k-mer models for computational efficiency and reliable signal simulation in nanopore sequencing applications.

Lifespan, characterized by significant interspecies variation, is shaped by an intricate interplay of genetic, environmental, and physiological factors. This study, anchored in Systems Biology, explores the influence of entropy, fractals, and Fibonacci energy optimization on lifespan evolution. Using the AnAge database, we elucidated the patterns linking the complex mechanics of aging with the Fibonacci sequence. Notably, our findings reveal a significant association between maximum lifespan and Fibonacci age, fostering an intriguing discourse on their connectedness. This connection sparks additional inquiries into the mechanisms and evolutionary significance of Fibonacci energy optimization in lifespan regulation. By weaving RNA structomics and DNA heritability considerations into a Systems Biology framework, we aim to unravel the sophisticated interactions bridging genotypes and phenotypes, offering fresh insights into the principles governing lifespan variation. Our study, synergizing Systems Biology principles with empirical examination, provides valuable insights into the multifaceted nature of lifespan and sets the foundation for future investigations in this fascinating field.

The intersection of genomic information and healthcare data presents a promising avenue for personalized medicine and improved patient outcomes. However, the integration of these diverse data sources poses significant challenges. We explore the imperative need for integrative frameworks that effectively harmonize genomic data with clinical records and other healthcare information. By bridging the gap between genomics and healthcare, such frameworks hold the potential to unlock insights into disease mechanisms, optimize treatment strategies, and enhance preventive care. Key considerations in developing these frameworks include interoperability, data privacy, standardization, and scalability. In this work we will present and discuss emerging methodologies and technologies aimed at addressing these challenges, including data integration techniques, machine learning algorithms, and secure data sharing platforms.

xxxx

Abstract. Transcriptomics-based benchmark dose (BMD) analysis is a promising approach for chemical risk assessment and toxicity testing. It provides mechanistic insights into chemical exposures and associated alerts. To enhance the implementation of toxicogenomics-based BMD analysis in the regulatory framework, the use of mechanistic evidence, such as the adverse outcome pathway (AOP) concept, is promoted.
In this work, we present BMDx 2, a new version of the BMDx tool for dose-dependent analysis of toxicogenomic data, which is currently under development. BMDx 2 allows the user to perform an AOP-based characterization of the mechanism of action of the tested chemicals by linking the dose-dependent genes with possible molecular initiating events and adverse outcomes.
The software efficiently handles complex BMD analysis of toxicogenomic data, including microarray and RNA-Seq data. BMDx 2 includes an expanded model list and enables consensus averaging of multiple models for each gene. Moreover, it implements a re-engineered code for easier extension to other models. With a comprehensive pipeline from data filtering to visualizing transcriptomic data and AOP-level BMD results, BMDx 2 enables efficient and robust analysis. Additionally, the tool offers visualization of AOPs categorized by various human health safety indicators, further enhancing its utility in hazard assessment.

Spatial omics techniques have made significant advancements in recent years. These techniques have enhanced our comprehension of the spatial localization of cells and cell-cell interactions. Cell-cell interactions are essential for maintaining tissue homeostasis. Dysregulation of these interactions can cause a variety of diseases, underscoring the importance of understanding their mechanisms. Ligand proteins, one of the signaling molecules responsible for cell-cell interactions, bind to cell surface receptors and trigger intracellular signaling pathways. The effective range of ligand signaling may play an important role in determining the occurrence of cell-cell interactions, but has not been systematically studied. In this presentation, we developed a new model of ligand spatial diffusion from spatial transcriptome data to estimate the spatial extent of ligand action. Specifically, using Visium data, we learned the diffusion distance as a parameter by regressing the expression of downstream target genes against the diffused ligand expression. This model succeeded in distinguishing between ligands with long and short effective distances, allowing analysis of cell-cell interactions taking into account the effective distance of the ligand. In this study, we present the results of our analysis using colorectal cancer specimens.

Next generation computational biology is needed to assist the emergent field of molecular system medicine. The development speed of omics tools fuels the increase of multi-omics knowledge databases, moving the medicine towards molecular grounds. The question is the integration of omics with physiology, for which new conceptual methods are in demand. Our work comes from the following observation: the bio-molecular relationships are extremely complex while the investigating methods, such as bio-networks and features extraction from bio-data sets, are too restrictive to model living organism complexity. Important dynamical information is lost between. The knowledge acquired until now is appealing to infer, using advanced mathematical objects, more biological insights, so better medical logic. With this vision we introduce a pioneering approach which address a central concept: the biological connectivity. Indeed, the network model, an engineering object, is too schematic to
describe connection in living system sense. We create a bio-compiler, as a comprehensive and flexible model, which main characteristic is ‘modus
communicatio’ as ‘modus operandi.’ Based on regular grammars, defined by means of bio-algebras having theorems/rules biological inputs, the
architecture of this compiler is created so that there is a unique minimal automaton accepting it. This construction is not only rigorous but it
produces a decidable state complexity. As an example of biocompilation architecture, we show how to asses breakpoints of the physiological
dynamic normalization between endocrino-immuno-lymphatic system, using communication between Endogrammar, Immunogrammar and
Lymphogrammar, associated language entities. Biocompilation opens rigorous and comprehensive perspectives on healthcare future, featuring
bio-connectivity as language communication.

The self-assembly of oligomers is essential for cellular differentiation and response to various stimuli. These processes are not random but result from intricate interplay within extensive molecular interaction networks. On the nanoscale, techniques like molecular dynamics (MD) simulations, including steered MD simulations and umbrella sampling approaches, can provide valuable insights into oligomer formation mechanisms. These techniques offer profound insights into the thermodynamics and kinetics of oligomer formation, providing a detailed view of the molecular forces and structural changes involved. On the microscale, studying the dynamics of Chemical Reaction Networks (CRNs) can help understand and predict oligomers’ holistic role in the behavior of more complex chemical systems such as large-scale protein-protein interaction networks. Furthermore, despite the challenges, studying the causal effects of these multiscale processes in harmony can offer a more comprehensive insight into protein assembly and interaction, contributing significantly to our understanding of cellular behavior and its regulation by molecular interactions. In this research, we present a multiscale modeling framework allowing the integrated study of nano and microscale biology of oligomers, exemplified using the PP1-GADD34-eIF2B complex that is critical in the termination of cellular stress response. We exploited Umbrella Sampling approaches to estimate the dissociation constants for each monomer and get structural insights into the PP1 holoenzyme complex assembly. Then, we leveraged CRN dynamics to assess the combinatorial complexity leading to the multimer organization and relating it to different cellular states.

Many representation formats for computational modeling, such as SBML and CellML, rely on markup languages and prioritize computer-friendliness, but often lack accessibility for human modelers in tasks like model curation, creation, verification, evaluation and expansion. There is a growing demand for a format that enables human users to preview and modify various aspects of models, including individual elements, interactions between elements, and the attributes associated with these interactions. In this work, we introduce the BioRECIPE (Biological system Representation for Evaluation, Curation, Interoperability, Preserving, and Execution) format to address this demand. BioRECIPE is a tabular format ideal for models structured as directed graphs, maintaining machine-readability and compatibility with various model development and analysis tools. BioRECIPE supports two distinct formats: event-based lists of interactions and element-based executive models. In the event-based format, each row corresponds to a single interaction, with column headers aligning with interaction attribute names, categorizing attributes into elements, interactions, context, and provenance. On the other hand, the element-based format assigns each row to a model element and supports a model’s static graph structure along with attributes necessary for dynamic analysis. This format also accommodates various schemes, ranging from less to more detailed, including static and dynamic parameters such as rate and timing. The BioRECIPE format is compatible with established formats like SBML and JSON, facilitating seamless translation between various output formats obtained from INDRA. It also serves as a standardized format within DySE framework and is integrated into many analysis tools, including FLUTE, VIOLIN, CLARINET, ACCORDION, FIDDLE, and PIANO.

Solid tumors have high interstitial fluid pressure (IFP), high mechanical stress, and low oxygen levels. Solid stresses may induce apoptosis, stimulate the invasiveness and metastasis of cancer cells and lower their proliferation rate, while oxygen concentration may affect the response of cancer cells to treatment. Although tumors grow in a nonhomogeneous environment, many existing theoretical models assume homogeneous growth and tissue has uniform mechanical properties. For example, the brain consists of three primary materials: white matter, gray matter, and Cerebrospinal fluid (CSF). Therefore, tissue inhomogeneity should be considered in the analysis.
This study established a physical model based on convection-diffusion equations and continuum mechanics principles. The model considers the geometrical inhomogeneity of the brain by including the three different matters in the analysis: white matter, gray matter, and CSF. The model also considers fluid-solid interaction and explicitly describes the effect of mechanical factors, e.g., solid stresses and IFP, chemical factors, e.g., oxygen concentration, and biological factors, e.g., cancer cell concentration, on growing tumors.
In this article, we applied the model on a brain tumor positioned within the white matter, considering the brain inhomogeneity to estimate solid stresses, IFP, the cancer cell concentration, oxygen concentration, and the deformation of the tissues within the neoplasm and the surrounding. Tumor size was estimated at different time points. This model might be clinically crucial for cancer detection and treatment planning by measuring mechanical stresses, IFP, and oxygen levels in the tissue.

The SysMod COSI organizes annual gatherings at ISMB. This short talk will introduce all speakers, organizers, and the main topics of the 2024 meeting. This year’s meeting incorporates three sessions covering, beginning with “Computational modeling of regulatory processes in Epithelial to Mesenchymal Transition (EMT),” followed by “Computational modeling approaches of immune responses and cellular proliferation”, and “Computational modeling of metabolic processes”. Two outstanding keynote speakers will present their visions on developments in these fields: Nathan Price from Thorne HealthTech, US, and Melissa Kemp from the Georgia Institute of Technology & Emory University. The event will close by awarding this year’s poster prizes. The event is hosted by Matteo Barberis and Meghna Verma.

Healthcare must become increasingly focused on extending healthspan and not only on treating disease after symptoms arise.  Indeed, this transition is essential to satisfactorily deal with the chronic diseases that account for the vast majority of healthcare costs today. To enable the precision health strategies of the future — what Lee Hood and I call The Age of Scientific Wellness (book, Harvard Press) — it is necessary to generate not only genomic data but also a large amount of longitudinal multi-omic data to quantify the health phenotype and to observe the earliest transitions to disease.  Such an approach enables predictive and preventive medicine. I will discuss how we have used such longitudinal ‘deep phenotyping’ data to: (1) map out the manifestation of genetic risk in the body, giving insight into intervention strategies to preemptively reduce disease risk on a personalized basis; (2) inform about how our gut microbiome and blood metabolites are related, and how the gut microbiome becomes more unique to each individual in healthy aging; (3) how the success of lifestyle/dietary-aimed interventions is quantitatively predicted by personal genetics, and (4) technological advancements to make gathering these data easier and cheaper for people, and (5) how digital twins provide new insights into deep personalization for brain health and new strategies for multi-modal clinical trials.  Taken together, such approaches help enable the future of health optimization and personalized, preventive medicine.

A major question in systems biology is to elucidate the gene regulatory mechanisms of cellular state transitions during developmental processes like cell differentiation and disease progression such as tumorigenesis. The advances in single-cell RNA-sequencing (scRNA-seq) technology has enabled an enhanced understanding of the dynamics of genome-wide gene expression. Yet, establishing gene regulatory networks driving cellular state transitions using scRNA-seq data remains challenging for a mechanistic understanding of cellular state transitions. Here, we introduce NetDes, a combined top-down bioinformatics and bottom-up systems biology approach, aimed at computationally generating ODE-based nonlinear dynamical models of core transcription factor regulatory circuits that recapitulate observed gene expression time trajectories. Our in-silico benchmarking demonstrates the advantage of NetDes in inferring the ground-truth regulators and their combinations. We applied NetDes to build the core regulatory circuit driving the differentiation of human iPSC to definitive endoderm using time series scRNA-seq data. The constructed gene circuit captures the regulatory interactions between stemness and Epithelial-Mesenchymal Transition (EMT) during this cell differentiation. Compared to existing network construction methods, NetDes has the advantage in capturing the gene expression dynamics during cellular state transitions using a single dynamical circuit model. Additionally, we performed systems biology simulations on the established ODE model to identify possible regulators and their combinations to drive the observed gene expression dynamics of the system. Our approach paves the way for a high-quality mechanistic modeling of the gene regulation of cellular state transitions.

Macrophages are immune cells which play a key role in the reaction to biomaterials. They exhibit a functional phenotype (or state) induced by the stimulus received and conditions of the microenvironment. This polarization process is governed by specific cytokines that are released by the macrophage itself, as well as produced by other cellular activation mechanisms. Cytokines act as phenotype markers within a heterogeneous range whose extremes are historically identified as pro-inflammatory or M1 and anti-inflammatory or M2. In such a context, this work aims to propose a predictive modeling approach for the simulation of the response to a pro-inflammatory stimulus in macrophages. This will allow us to subsequently simulate the immune reaction induced by the presence of biomaterials at the cellular level, with the final goal to build a digital twin of the inflammatory response in a foreign body reaction. To do that, existing Ordinary Differential Equation (ODE) and Agent Based (AB) models have been considered and validated with in-vitro experimental data. Preliminary results highlight a better agreement of the AB approach over the ODE models taken into account in this work. This specific scheme is making simplified assumptions on spatial resolution and diffusion of inflammation (both cytokine and macrophages). However, the good agreement that we have observed in this simplified model encourages the use of a more advanced and comprehensive hybrid simulation platform based on AB modeling which implements a more thorough description of the intracellular pathways and the microenvironment.

Adaptive stress response networks (SRNs) are invoked when chemical exposures induce DNA damage, oxidative stress, unfolded proteins, hypoxia, or heat shock and are essential for maintaining cell health. With a highly conserved architecture for sensing and countering cellular stress, SRNs are also pivotal in activating senescence, apoptosis, and autophagy pathways if stress cannot be resolved. Perturbing SRNs beyond some threshold tips cells over from adaptive to adverse phenotypes. We are investigating these critical “”tipping points”” using a combined approach of literature mining, computational modeling, and high-throughput data analysis. We aim to elucidate how SRN dynamics dictate cellular phenotypes and propose Boolean Networks (BNs) to identify these tipping points by: 1) Constructing a biological knowledge graph (KG) of chemical stress inducers; 2) Utilizing the KG to build BNs and simulate SRNs; and, 3) Validating predictions with transcriptomic data from HepaRG cells. Herein, we describe the KG of 500 chemical stress inducers for which we developed BNs to simulate dynamic cell trajectories in DNA damage response for drugs and chemicals. By elucidating the molecular determinants of tipping points, this research furthers our understanding of cellular resilience in health and toxicity. This abstract does not reflect US EPA policy.

The eukaryotic cell cycle is driven by waves of cyclin-dependent kinase (cyclin/Cdk) activities that rise and fall with a timely pattern called “waves of cyclins”. This pattern guarantees coordination and alternation of DNA synthesis with cell division. Through computational modelling, we have recently identified a minimal network underlying cyclin/Cdk1 autonomous oscillations in budding yeast that we name the Yeast Minimal Cell Cycle. Here, we first explore by ‘learning from building’ whether these cell cycle oscillations may be achieved by synthesizing a functional genome consisting of a minimal set of cell cycle genes. We consider nine genes involved in the waves of cyclins: G1 cyclins (CLN1,2), mitotic cyclins (CLB1-6), and their positive and negative regulators (FKH1,2 and SIC1, respectively). Selected genes were pairwise deleted by CRISPR from their native loci in the yeast genome and simultaneously relocated with their native promoters into a synthetic gene cluster in the same cell. We then remove combinations of genes from this cluster. The frequency of gene loss and the growth rates of these strains were analysed and compared to kinetic models of the cyclin/Cdk network that are verified against quantitative data of Clb dynamics. We unravel a novel molecular design that synchronizes Clb/Cdk1 oscillations. Through integration of Synthetic and Systems Biology, this work shows that a minimal set of genes of the Yeast Minimal Cell Cycle can reproduce cell cycle oscillations, indicating that the genetic complexity of the yeast cell cycle can be reduced to identify a novel molecular network underlying cell proliferation dynamics.

Understanding cell-cell interactions (CCIs) is essential yet challenging due to the inherent intricacy and diversity of cellular dynamics. Existing approaches often analyze global patterns of CCIs using statistical frameworks, missing the nuances of individual cell behavior due to their focus on aggregate data. This makes them insensitive in complex environments where the detailed dynamics of cell interactions matter. We introduce CellAgentChat, an agent-based model (ABM) designed to decipher CCIs from single-cell RNA sequencing and spatial transcriptomics data. This approach models biological systems as collections of autonomous agents governed by biologically inspired principles and rules. Validated against seven diverse single-cell datasets, CellAgentChat demonstrates its effectiveness in detecting intricate signaling events across different cell populations. Moreover, CellAgentChat offers the ability to generate animated visualizations of single-cell interactions and provides flexibility in modifying agent behavior rules, facilitating thorough exploration of both close and distant cellular communications. Furthermore, CellAgentChat leverages ABM features to enable intuitive in silico perturbations via agent rule modifications, pioneering new avenues for innovative intervention strategies. This ABM method empowers an in-depth understanding of cellular signaling interactions across various biological contexts, thereby enhancing in-silico studies for cellular communication-based therapies.

Cell-type transitions, crucial for processes including cell quiescence, cell cycle, and embryogenesis, involve intricate metabolic rewiring to optimize competing biological objectives. The competition for cellular resources is often explored through the lens of Pareto optimality and metabolic trade-offs. Despite advancements in understanding these dynamics in unicellular organisms, the metabolic trade-offs in multicellular systems, especially during embryonic cell-state transitions, remain largely unexplored. Addressing this gap, we introduce the Single Cell Optimization Objective and Tradeoff Inference (SCOOTI) framework, a novel computational approach grounded in optimization theory, designed to infer cell-specific metabolic objectives from omics data. By integrating gene expression, protein abundance, and metabolite concentration data with genome-scale metabolic models, SCOOTI leveraged meta-learner regressors to quantitatively elucidate the metabolic objectives underlying cell quiescence, proliferation, and embryogenesis. Our analysis reveals distinct metabolic objectives across cell-cycle phases and embryonic development stages, highlighting the role of specific metabolites in driving these transitions. Notably, the framework uncovers a shift from a high entropy, multitasking metabolic system in early embryogenesis to a more deterministic metabolic focus on biomass production and cell growth in later stages. This shift is exemplified by the trade-offs between glutathione-mediated redox balance and biomass precursor synthesis, suggesting a Pareto optimality scenario where balancing redox status and growth-related objectives is crucial for embryonic development. Our findings challenge the traditional biomass maximization model, proposing instead that cellular metabolic objectives are highly context-dependent, varying significantly between quiescent and proliferative states, and across developmental stages.

Bacillus subtilis is a key organism in biotechnology, with its metabolic capabilities offering potential for various valuable products like levan. Levan, a fructose polymer, has diverse applications such as the formulation of hydrogels, drug delivery, and wound healing, among others. But B. subtilis’ metabolic models till date are not that well-annotated as compared to other model organisms’ metabolic models and the exploitation of an organism’s metabolic capability requires an improved model with expanded gene coverage. So, this study aims to enhance the metabolic model of B. subtilis for improving levan biosynthesis, addressing the need for accurate prediction of metabolic outcomes in industrial applications. We used structural systems biology technique, which offers a promising approach to enhance predictive power, for the refinement of the levan biosynthesis pathway in B. subtilis. Using AlphaFold2, structural models were generated for critical genes lacking full-length PDB structures, which were then integrated into the model. Findings elucidate previously overlooked structural aspects of levan biosynthesis, while leveraging the STRING database to optimize product yield. We provide a detailed understanding of the levan biosynthesis in B. subtilis, shedding light on previously overlooked structural aspects of a pathway. This metabolic model acts as an input to further set of applications in advancing metabolic engineering efforts.

Engineering multicellular systems is enhanced by understanding how collective organization arises during developmental processes through mechanical, biochemical and electrical communication. Which aspects of these processes can be circumvented, accelerated or modified according to specification to yield robust, reproducible organoids? Computational models that simulate the growth, division, and differentiation of pluripotent cells into emergent structures could accelerate experimental design, yet currently lag in their ability to inform organoid culture protocols. I will discuss my lab’s computational results from developing agent-based models that capture heterogeneity and stochasticity within colonies and aggregates to both i) formulate hypotheses of intercellular communication during stem cell differentiation and ii) design new organoid structures using synthetic biology components. To address the challenges of agent-based model optimization, we have pursued new methods for analyzing microscopy images and simulation results by topological data analysis. Through a tight iteration between computation and experimentation, we established a critical role of intercellular transport, adhesion, and cell cycle asynchrony in the propagation of dynamic patterning in engineered iPSC systems.

This remark will provide a brief overview of the SysMod meeting, covering the speakers, chairpersons, and organizers. The event consisted of four sessions that delved into diverse methodologies and applications of computational systems modeling in the field of biology. We were privileged to have two distinguished keynote speakers, Nathan Price and Melissa Kemp, who delivered insightful talks. We express our gratitude for all the valuable contributions made to the scientific posters and we recognize and honor the best ones through the annual SysMod poster awards in 2024.